Metadata is everything attached to your music that is not the audio file itself.
Title, artist name, contributors, genre, language, release date, identifiers, rights, territory, explicit flag, cover art. All of it. The audio is the cargo. The metadata is the shipping label, and the shipping label is what decides whether you get paid.
This guide is for any artist or label-side operator who has ever stared at an intake form and wondered which fields actually matter.
What is music metadata?
Metadata is the structured description of a recording and the release it sits on. It comes in roughly four tiers.
- Descriptive metadata — title, artist, featured artists, contributors, genre, mood, language of lyrics. The stuff a listener sees.
- Administrative metadata — ISRC, UPC, ISWC, label, copyright (C and P lines), release date, original release date. The stuff that drives royalty matching.
- Technical metadata — audio codec, bit depth, sample rate, channels, file format. The stuff that decides whether the DSP accepts the asset.
- Rights metadata — territory by territory, who owns the master and the composition, what commercial models are licensed (stream, download, ad-supported, premium).
All four tiers ride into the DSP inside a single DDEX ERN message. Get any tier wrong and something breaks.
Why does metadata exist?
Because a Spotify catalog of 130 million tracks cannot be navigated by ear. Search, recommendation, playlist eligibility, royalty matching, chart compilation, sync clearance, neighbouring-rights collection, language-specific store placement, age-gating, every single one of these depends on a database query against metadata, not against the audio.
When a fan types “Burna Boy live in Lagos” into Spotify search, no part of that query touches the audio. It is all metadata. When SoundExchange pays a master recording royalty for a US webcast, the matching engine reads the ISRC and contributor data from the metadata. The audio is irrelevant to the payment.
This is why clean metadata is the single highest-leverage thing an indie artist can control. It costs nothing and it compounds for the life of the catalog.
How does metadata work in practice?
The journey of a single field, take the producer credit on an Afrobeats single:
- The artist types “P. Priime” into the producer field on the distributor’s upload form.
- The distributor maps that field to a DDEX Contributor element with role code `Producer`.
- The ERN ships to Spotify, Apple Music, Boomplay, Audiomack, etc.
- Each DSP ingests, indexes, and displays the credit in their UI (Spotify Credits, Apple Music Credits, etc).
- When a sync agency later searches for “tracks produced by P. Priime,” they query against that contributor field.
- When a neighbouring-rights society in Europe pays out for a non-featured performer, they match on the same field.
One field. Six downstream consumers. Get it wrong and six things degrade silently.
What metadata means for indie artists and labels
Three working rules.
Decide the canonical artist name before you release anything. “Tems,” “TEMS,” “Tems (Nigeria),” and “Tems & Wizkid” are four different artists to a DSP matching engine. Pick one capitalisation, one spelling, one disambiguation pattern, and use it on every release for the rest of your career. Distribution dashboards expose an Artist ID (Spotify URI, Apple Music ID) that you must paste into every new release to keep the catalog merged.
Featured artists go in the featured-artist field, not in the title. “Song Title (feat. X)” is rendered by the DSP from the structured featured-artist relationship. If you stuff the feature into the title string, the featured artist never gets the credit on their profile and the song will not show on their discography.
Cover art is metadata too. 3000×3000 JPG, sRGB color space, no embedded text outside the title and artist name, no DSP logos, no social handles. Apple Music in particular will reject otherwise.
Common metadata mistakes and gotchas
- Genre stuffed with five tags. Most DSPs accept one or two. Pick the dominant one. Extra tags do not increase discoverability; they confuse the recommender.
- Original release date set to today on a re-release. A 2018 song re-released in 2026 with today’s date loses all historical signals at Spotify and Apple. Always set original release date to the actual first release.
- Language field set to “English” on a Yoruba-heavy track. This affects regional store placement, search routing, and lyric-display partners like Musixmatch. Set the dominant lyric language, not the title language.
- C-line and P-line confused. C (©) is the composition copyright owner. P (℗) is the sound recording copyright owner. They are usually different entities. Both must be on the release.
- Explicit flag wrong. If your track has profanity and the flag is false, the DSP can pull the release on a complaint and ban the account from re-uploading for a period.
- Contributor names without IPI numbers. Particularly for writers, an unlinked contributor name floats free and matches nothing at the PRO.
- Two versions of a release with identical metadata. Distributor strips the duplicate or rejects both. Differentiate or consolidate.
How InterSpace Distribution handles this
InterSpace Distribution is a global distributor in the same category as DistroKid, TuneCore, ONErpm, Symphonic, EMPIRE, and Believe. We validate every metadata field against DSP-specific rules before delivery, expose contributor roles that map cleanly to DDEX, preserve original release dates through migrations, and flag duplicate-catalog risk at upload time. Get started at cms.interspacemusic.com/signup.